301 results about "Fuzz testing" patented technology

The invention discloses an industrial control protocol intelligent fuzzy test method and system based on a multi-agent large model, and relates to an industrial information security test method and system.According to the industrial control protocol intelligent fuzzy test method and system based on the multi-agent large model, the large model is enhanced by integrating the field of the retrieval enhancement generation (RAG) and the low-rank quantization adaptation (QLoRA) technology, and a multi-agent cooperation mechanism is combined; and full-automatic, high-precision and end-to-end fuzzy testing of the industrial protocol is realized. The system generates seed data meeting protocol specifications through the seed generation agent, the test case generation agent executes field, structure and semantic variation based on the seed data, efficient test cases are dynamically generated, a test strategy is optimized in real time through the feedback analysis and strategy adjustment agent, and the vulnerability discovery capability is improved. By dynamically adjusting the variation strategy and optimizing the generation of the test case, the coverage, the accuracy and the efficiency of the test are improved, and the security and the vulnerability detection capability of the industrial control system are remarkably improved.

The invention discloses a knowledge extraction and envelope coverage-oriented software vulnerability test method and related equipment, and the method comprises the steps: obtaining vulnerability key information of target software, and obtaining sensitive function features through defect dependence analysis and construction based on the vulnerability key information; performing semantic analysis on the analysis report and the code abstract associated with the sensitive function characteristics, screening to obtain a target sensitive function, performing path envelope reverse tracking on the target sensitive function, and constructing to obtain a target path envelope; and obtaining coverage information enveloped by the target path, and carrying out fuzzy testing on the basis of the coverage information in combination with the iteratively optimized variation sample. According to the method, through intelligent closed loop of analysis-positioning-testing-feedback-optimization, static analysis provides accurate guidance for dynamic testing, and the static analysis strategy is inversely optimized by the result of the dynamic testing, so that the maximum improvement of the testing efficiency and the vulnerability discovery accuracy is realized in limited testing resources, and the testing efficiency and the vulnerability discovery accuracy are improved. The method can be widely applied to the technical field of software security.

The invention discloses a stateful protocol fuzz testing method and system based on a large model construction state machine, and the method comprises the steps: firstly preprocessing a protocol specification document, guiding an information large model to analyze through cue words to obtain state machine information, constructing an initial state machine through the state machine large model, and initializing a message dictionary; secondly, protocol testing is carried out based on an initial state machine, a prompt word is input into a large testing model to understand a flow log, a message field supplement and improvement dictionary is obtained, message request fields and response codes in the message dictionary are combined, the initial state machine is improved, and a fine-grained state machine is obtained; and finally, on the basis of a fine-grained state machine, inputting cue words, analyzing a blind area covered by the current test by an analysis large model, and guiding a fuzzy tester to preferentially test an uncovered state which is most likely to trigger the deep vulnerability, thereby completing the fuzzy test. According to the method, the fuzzy device directionally tests the potential test state area, and accurate and efficient fuzzy testing is carried out.

The invention provides a protocol security fuzzy test system and method based on a speed reduction bridge, the speed reduction bridge is integrated on an FPGA board card, and the speed reduction bridge comprises a protocol fuzzy test processing module, a clock domain crossing buffer and flow control module and a programmable time sequence fault injection circuit. The protocol-based fuzzy test processing module is used for detecting feature information and state information of a data stream flowing through the speed reduction bridge and outputting an action instruction; and based on different action instruction types, driving different modules to inject different types of abnormities into the data stream, and realizing fuzzy testing based on the abnormal data stream. The fuzzy test processing module is embedded in the FPGA logic, different rules can be matched according to the feature information of the data flow, different abnormal injection of the data flow is achieved, the abnormal injection opportunity is determined according to the link state information, the speed reduction bridge has the dual functions of normal protocol relaying and active abnormal injection at the same time, and the reliability of the speed reduction bridge is improved. The test of various abnormities of the equipment to be tested can be realized.

The invention discloses an MQTT protocol fuzzy testing method and system based on a large language model guiding sequence generative adversarial network, relates to the field of fuzzy testing, effectively combines the advantages of the large language model and the sequence generative adversarial network, and analyzes basic training data by introducing an intelligent analysis technology of the large language model. And the quality of the training sample of the sequence generative adversarial network is obviously improved. According to the method, the initial seeds can be enriched through the large language model, the discriminator can be assisted to perform multi-dimensional evaluation on the data generated by the generator, and the authenticity and diversity of the test seeds are enhanced. The generation of redundant test cases is greatly reduced, so that the fuzzy test process is more efficient. Meanwhile, a seed elimination technology based on an energy value is designed, it is further ensured that the tested seeds have higher quality, and the efficiency of vulnerability discovery is effectively improved. Meanwhile, according to the maximum coverage length, the test strategy is dynamically adjusted, continuous optimization of the test process is achieved, and a solid guarantee is provided for the test. According to the method, multi-level variation is carried out on the test seeds by utilizing a large language model, vulnerability information existing in the test seeds is quickly positioned, and meanwhile, multi-dimensional and deeper mutation is carried out on the test seeds, so that more existing vulnerability information is explored, namely, the multi-level mutation is carried out on the test seeds; the test seeds which successfully cause the vulnerabilities are verified and optimized by means of the response of the test target, the success and failure positions are summarized, and a more reliable guarantee is provided for subsequent variation.

An example method can include, obtaining information about a machine learning (ML) model configured to detect malicious activity on a computer system, wherein the information includes one or more of a model type of the ML model, an output type of the ML model, or a type of malicious activity that the model is trained to detect, receiving a training dataset, wherein the training dataset includes a plurality of unlabeled examples, generating an additional dataset based on the training dataset using a generative model, wherein the additional dataset includes a plurality of additional unlabeled examples, and, training the machine learning model to generate labels for each example in the training dataset and the additional dataset, using a combination of the training dataset and the additional dataset, wherein the training includes adjusting one or more parameters of the machine learning model based on accuracy of the generated labels.

The invention discloses a fuzzy test verification method and system for a password service interface, and belongs to the technical field of password service technologies and software testing. The method comprises the following steps: firstly, constructing a password service interface feature library, and extracting key feature information; designing a multi-dimensional fuzzy test variation strategy based on the password service interface feature library, and generating a fuzzy test case set covering grammar exception, semantic exception and encryption scene exception; realizing communication adaptation between a test case and a target password service interface through a dynamic adaptation module, executing a fuzzy test, and collecting information such as response data and an operation state of the interface in real time; and finally, carrying out deep analysis on the acquired information through an anomaly detection and analysis module, positioning a vulnerability type and a trigger path, and generating a standardized test report. According to the method, different types of password service interfaces can be accurately adapted, the test coverage and the vulnerability discovery efficiency are remarkably improved, and reliable support is provided for security verification of the password service interfaces.

The invention relates to the cross technical field of artificial intelligence and software security testing, in particular to a deep learning framework fuzz testing method based on large model cue word optimization, which is used for improving vulnerability mining efficiency and testing intelligence level of fuzz testing on a deep learning framework. According to the method, the advantages of a large language model in the aspects of code understanding and generation are fully utilized, and efficient vulnerability detection of a deep learning framework is realized by introducing a cue word adaptive optimization and variation mechanism. The method mainly comprises the following steps: (1) providing a deep learning framework API classification method and a cue word routing mechanism; (2) proposing a large model cue word adaptive optimization mechanism; and (3) proposing a deep learning framework fuzzy test variation strategy and a dynamic selection mechanism. According to the method, the automation and vulnerability discovery capability of fuzzy testing can be remarkably improved while the generation quality is ensured, and the method has relatively high universality and application value.

The invention discloses a static binary code taint analysis method based on a propagation action scope, and relates to the field of static binary code taint analysis, and the static binary code taint analysis method comprises the following steps: extracting a program instruction set and a control flow structure based on a disassembling result of a target binary code; calculating a value set with a source of the binary code based on the control flow graph; on the basis of the control flow diagram and the value set with the source, executing cross-function stain propagation analysis, and identifying memory positions or registers influenced by pollution in each function and propagation action ranges of the memory positions or the registers; and extracting all instruction sets using the taint data based on the taint and the propagation action range thereof. According to the method, the false alarm rate of static binary taint analysis can be reduced and the instruction set involved in the taint analysis can be reduced without increasing the analysis overhead, so that the method has important significance in improving the instrumentation efficiency and the operation efficiency of dynamic taint analysis and improving the accuracy of protocol reversion, fuzzy testing and vulnerability mining based on the taint analysis.

The embodiment of the invention discloses an intelligent fuzzy testing method and device based on a large language model and a medium, and relates to the technical field of fuzzy testing, and the method comprises the steps: obtaining and analyzing a source code of a target code library to construct a structured knowledge graph, and evaluating a function vulnerability risk through the large language model and the structured knowledge graph, generating a test target sorting list; performing multi-dimensional code analysis on the selected test targets in the test target sorting list to generate seed input so as to construct a compilable test driver through the seed input; performing a test driver in an isolated environment, performing a fuzz test using a seed input, collecting a crash report to analyze the crash report, identify a crash cause, and generate a minimization replicator; and generating a candidate patch based on the crash reason, and verifying the validity of the candidate patch through a minimization reproducing device and a structured knowledge graph so as to realize the fuzzy test of the source code of the target code library.

A method for testing device software of a device using a fuzzing algorithm. The method includes: initializing the fuzzing algorithm to detect an initial behavior of the device software of the device under test; executing a fuzzing test loop including: generating a fuzzed message from predefined message types; sending the generated, fuzzed message to the device under test to test the device software on the basis of the fuzzed message; detecting side-channel information during testing of the device software on the basis of the fuzzed message; recognizing anomalies in the side-channel information using at least one machine learning model and / or statistical algorithm; and if an anomaly is recognized, adjusting a parameter including adjusting fuzzability weights, of the initialized fuzzing algorithm and performing a next loop iteration of the fuzzing test loop; and if no anomaly is recognized, performing the next loop iteration of the fuzzing test loop.

The embodiments of this disclosure provide a kernel fuzzing method and apparatus based on unicorn simulation. The method includes: creating a unicorn engine in the harness of unicorefuzz, mapping all required memory regions and setting the initial values ​​of each register according to the needs of the target program; passing the fuzzing input data generated by AFL to the unicorn engine; after the user specifies the target function address and breakpoints for the fuzzing, the unicorn engine starts the AFL fork server to simulate the execution of the fuzzing of the target program; and loading memory as needed and handling abnormal situations of the target program during the fuzzing process.

The invention discloses a semantic perception kernel fuzzy test method and device based on a large language model, and the method achieves the structural and context-related semantic analysis of a system call sequence in a variation stage through the introduction of the large language model with the language understanding and code modeling capabilities. And a variation strategy with logic consistency and attack potential is dynamically generated, so that the effectiveness and diversity of fuzzy test input are greatly improved. Specifically, semantic modeling and structured understanding are performed on a system call sequence by using a large language model, a variation strategy instruction with context relevance and execution reasonability is generated, and fuzzy test path exploration is continuously optimized through a dynamic feedback mechanism. The defects of a traditional fuzzy test method in variation efficiency, coverage depth and complex state adaptation are effectively overcome, and a more intelligent, self-adaptive and efficient vulnerability detection and security guarantee solution is provided for an operating system kernel.

The application discloses a concurrent program vulnerability detection method based on fuzzy testing and static analysis, and comprises the following steps: compiling a to-be-detected program source code, obtaining a bytecode file according to the compiling, then performing static analysis to obtain a concurrent vulnerability pair; performing program insertion on the concurrent vulnerability pair to obtain a program binary file with inserted code; taking the program binary file with inserted code as a test program, adding an initial input file in an input queue, adding a concurrent vulnerability-oriented mechanism in AFL and adding a judgment mode of whether an input is interesting, and pruning unfeasible paths; performing fuzzy testing to realize vulnerability detection. The application can not only utilize the characteristics of concurrent program vulnerabilities, use an optimized redundant node static analysis technology to reduce the number of concurrent vulnerability detections, but also use a fuzzy testing technology to efficiently reproduce the occurrence process of the concurrent program vulnerabilities, so that more efficient and more intuitive concurrent vulnerability detection is realized.

The application discloses an industrial control system protocol fuzzing test method based on state conversion, and particularly, the communication flow of an upper computer and a PLC is analyzed, a protocol finite state machine of a communication interaction process is constructed, and the PLC is subjected to the fuzzing test by using test cases violating the finite state machine in a test process, so that the purpose of triggering unknown vulnerabilities of the PLC is finally achieved. The application uses the protocol finite state machine to guide the state conversion path of the fuzzing test process. The method for obtaining the test cases violating the original protocol finite state machine and then performing the fuzzing test on the equipment can efficiently find vulnerabilities. The confirmation method for the state conversion leading to the vulnerabilities can obtain the shortest state conversion path leading to the vulnerabilities.

The application discloses a kind of stateful network protocol-oriented fuzz testing method, constructs the corpus of initial test case, and carries out compiling and inserting plug-in operation;Through state selection algorithm, the target state of each round of fuzz testing is selected;From the test case set, the corresponding test case queue is selected, then a test case is selected from the test case queue to generate new test case by variation operation, then the new test case is sent to network protocol server to carry out fuzz testing;After a round of fuzz testing is completed in network protocol server, the feedback information of network protocol server is collected, including code coverage, state coverage and state transition, and subsequent fuzz testing process is guided according to feedback information.The above-mentioned method can solve the problems existing in state selection, test case scheduling and other aspects of traditional network protocol fuzz testing, improve network protocol fuzz testing efficiency and vulnerability mining capability.

The AFDX network protocol fuzzing method of the application is suitable for testing of an airborne network protocol, comprising: S1, taking configurable parameters in the AFDX protocol as input of the test data packet, wherein the configurable parameters comprise virtual link configuration, bandwidth configuration, time scheduling configuration and protocol frame field configuration; S2, calling the test data packet to test the network protocol; S3, sending the test data packet to the tested device through the AFDX network; S4, obtaining the test result of the tested device; and S5, based on the test result, marking the test case causing an abnormal scenario as an effective test case by using artificial or automatic means, thereby improving the test efficiency of the network protocol and related software.

The invention discloses a description language tool for a protocol fuzz test, and the tool comprises a PDL module which is used for defining a protocol message structure, a session state, and association between fields; the analysis module is used for analyzing the PDL definition into an intermediate representation; the fuzzy test engine is used for generating a use case by adopting a three-stage coding assembly line according to the intermediate representation and the session context; and the execution module is used for sending the use case and receiving the response. The PDL of the tool supports bit-level precision, a composite structure and an inheritance mechanism, a semantic perception variator is arranged in the fuzzy test engine, and consistency of use case generation and response analysis is achieved through single-definition bidirectional semantics. Accurate modeling and dynamic simulation of a complex protocol are realized through a unified description language, and the efficiency, the coverage rate and the maintainability of fuzzy testing are remarkably improved.

The invention relates to a CAN bus fuzzy test method and device, computer equipment and a storage medium. The method comprises the steps that an original CAN message is acquired, the original CAN message is coded through a pre-trained generator, a coding result is decoded, a reconstructed sequence is obtained, the generator is obtained through sequence-to-sequence model training, the reconstructed sequence is distinguished through a generative adversarial network discriminator according to the original CAN message, and the reconstructed sequence is obtained. And determining the abnormal probability of the reconstructed sequence, adjusting the reconstructed sequence according to the abnormal probability, determining a test message according to the adjusted reconstructed sequence, and injecting the test message into the CAN bus to perform the fuzzy test of the CAN bus. By adopting the method, the test message with high coverage rate and pertinence can be generated, so that the test depth and the test effect are improved.

The present invention discloses a VxWorks transport layer protocol fuzz testing method based on feature tracking. This method performs consistency checks on the message response flag features of the transport layer TCP protocol of the closed-source operating system VxWorks, thereby driving seed iterative mutation and dynamically guiding the fuzz testing process. A progressive fuzzification method that fuses feature relationships between messages is used to construct test cases with high acceptance rates and high complexity. A dual anomaly detection mechanism that combines client connection detection with system anomaly task detection based on WDB RPC is used to determine and detect abnormal states during the fuzz testing process. The present invention implements a lightweight, practical, and efficient fuzz testing method for the TCP protocol of the VxWorks operating system.

A coverage-guided code generation and fuzzing method based on a large model is proposed to address the problems of existing large model fuzzers, which not only ignore the guiding role of coverage information in the fuzzing loop but also generate prompts too hastily. This invention comprises four stages: prompt generation, fuzzing loop, mutation operator selection, and vulnerability detection. This method proposes a prompt generation method based on expert templates and a fuzzing loop method guided by unequal interval coverage information. The former lowers the barrier to entry for users of CoverFuzz and improves the quality of initial prompts; the latter, guided by coverage information, better covers the object under test to detect vulnerabilities. This invention significantly improves the effectiveness of programs generated by fuzzing based on large models, solves the problem of existing large model-based fuzzers ignoring coverage information, and also achieves higher code coverage.

The invention provides a compiler fuzz testing technology for generating complex compiling options based on a large model. According to the technology, the large model is effectively introduced into each stage of compiler testing. Firstly, in the aspect of a test code data set, a test code is obtained in combination with a historical defect report of a compiler, and a code snippet embedded in a natural language text of the historical defect report is obtained by using a large model; in order to more efficiently and specifically combine the complex compiling option test compiler, the technology combines a large model cue word project to generate a more targeted complex compiling option test compiler for each section of test program. Meanwhile, in order to avoid the situation that input codes exceed the context length limit of a large model, the technology abstracts source codes on the intermediate code level. The method has the main advantages that the code understanding and generating capacity of the language model is well utilized, legal and diversified complex compiling option input programs are automatically generated, and therefore the testing efficiency of the compiler is improved.

The invention discloses a Solana chain program transaction sequence dependence defect detection method based on fuzzy testing and symbolic execution, and the method takes an ELF executable file of a Solana chain program as an input, and takes a detected transaction sequence dependence defect report as an output; in order to detect the transaction sequence dependency defect in the program on the Solana chain, the method is based on the thought of fuzzy testing and symbolic execution, the program on the input Solana chain is tested and analyzed, an initial block chain simulation state and a seed transaction are generated from an executable file of the program, and then loop testing is carried out. And when the test is finished, outputting detailed reports of all defects found in the process. The method provided by the invention has the advantages of effectiveness and high efficiency, and can more effectively discover the transaction sequence dependency defect in the Solana chain program.

The invention provides a differential fuzz testing method for a JIT compiler in a JS engine and terminal equipment. The differential fuzz testing method comprises the following steps: constructing a plurality of initial test cases which comprise JIT compiler error inducing elements and can simultaneously trigger a byte code interpreter optimization function and a JIT compiler optimization function; fixing a return value of the dynamic function in each initial test case; fixing an execution context based on a strict mode, and isolating a global object; executing the test case for multiple times by utilizing a byte code interpreter, and performing consistency detection on an interpretation execution result obtained after multiple times of execution; and triggering an optimization function of the JIT compiler for the consistent test cases by utilizing a consistency detection result, and detecting whether the JIT compiler has an error or not according to execution results before and after optimization. According to the invention, the accuracy of JIT compiler error detection can be improved.

The application discloses a fuzzy test seed screening method and device. The method is based on multiple sets of fuzzy test seeds. Seed messages corresponding to the fuzzy test seeds are sent to a test object in sequence, and the change degree of each state feature of the test object when processing the seed messages is monitored. The change degree of the state of the test object when processing the messages can determine the complexity of the device under test when processing the messages or commands. The more complex the processing logic is, the more likely bugs exist, and the test should be focused on. Finally, the fuzzy test seeds are sorted according to the state influence degree of the seed messages on the test object, so as to determine the screening priority of each fuzzy test seed, so that the fuzzy test seeds with higher priority are screened preferentially to generate test cases, thereby improving the test efficiency.

A security inspection device (100, 100A, 200, 300) comprises: a switching unit (111) that switches between enabling and disabling of a vulnerability mitigation function (specific function; SSP and ASLR) issued by an OS (120) to a target program (140); a search unit (112) that, in a state in which the mitigation function for vulnerabilities of the target program (140) is enabled, employs fuzzing to search for input information with which an output differs each time the target program (140) is executed, despite the input information being the same; and a confirmation unit (113) that, when the input information found by the search unit (112) is input to the target program (140) in a state in which the mitigation function is disabled, confirms whether or not randomness remains such that the output differs each time the target program (140) is executed.

This invention discloses an IoT fuzzing method based on LLM-guided and FSM dynamic inference, belonging to the field of IoT network security and software testing technology. Addressing the problems of low coverage and inaccurate state machine inference in current IoT protocol fuzzing, this invention first constructs an initial FSM by combining IoT protocol specifications and captured traffic data. Then, it generates a large number of test cases through mutation of seed test cases for fuzzing testing. Features are extracted from device responses, and state identification is performed by calculating similarity. When a new state appears, the FSM and state fingerprint database are updated. When coverage becomes a bottleneck, LLM-guided path inference is used to generate extended sub-FSMs and test cases, which are then executed. The FSM is then corrected based on the test results. This invention enables high-precision automated construction of IoT protocol state FSMs, improving test coverage and enhancing the efficiency and accuracy of vulnerability discovery.

This application proposes a fuzzing method and system for CXL devices based on hierarchical state abstraction. While accurately capturing cross-layer state changes in CXL to effectively discover vulnerabilities, it reduces the amount of state data through hierarchical feature selection, reduces deduplication overhead through incremental hashing, and reuses intermediate state operation sequences through a shared state pool, thereby shortening the average execution time of each test case and improving the overall throughput of fuzzing. This application significantly shortens the average test cycle while maintaining a high vulnerability detection rate, especially demonstrating a significant improvement in coverage efficiency for multi-hop memory consistency protocol scenarios.

This invention discloses a code defect analysis and repair method and system based on a large language model. The method includes: receiving crash information triggered by fuzzing or program execution, and performing crash analysis to obtain crash analysis results; obtaining key defect analysis information based on the crash analysis results, and constructing structured defect analysis information based on the key defect analysis information; inputting the structured defect analysis information into a defect analysis model, and performing reasoning analysis on the code defects according to a three-round reasoning strategy to obtain defect analysis results; searching a historical repair database based on the defect analysis results to obtain historical repair retrieval results, and inputting the defect analysis results and historical repair retrieval results into a patch generation model to generate candidate patch code; performing sandbox verification on the candidate patch code to obtain valid patch code and patch verification results, and outputting an analysis and repair report based on the defect analysis results, valid patch code, and patch verification results.